UvA-DARE is a service provided by the library of the University of Amsterdam (https://dare.uva.nl)
Filling the gaps
The endothelium in regulating vascular leakage and leukocyte extravasation
Schimmel, L.
Publication date
2018
Document Version
Other version
License
Other
Link to publication
Citation for published version (APA):
Schimmel, L. (2018). Filling the gaps: The endothelium in regulating vascular leakage and
leukocyte extravasation.
General rights
It is not permitted to download or to forward/distribute the text or part of it without the consent of the author(s) and/or copyright holder(s), other than for strictly personal, individual use, unless the work is under an open content license (like Creative Commons).
Disclaimer/Complaints regulations
If you believe that digital publication of certain material infringes any of your rights or (privacy) interests, please let the Library know, stating your reasons. In case of a legitimate complaint, the Library will make the material inaccessible and/or remove it from the website. Please Ask the Library: https://uba.uva.nl/en/contact, or a letter to: Library of the University of Amsterdam, Secretariat, Singel 425, 1012 WP Amsterdam, The Netherlands. You will be contacted as soon as possible.
1Department of Molecular Cell Biology, Sanquin Research and Landsteiner Laboratory, Academic
Medical Centre, University of Amsterdam, 1066CX Amsterdam, The Netherlands. 2Genetics and
De-velopmental Biology, Center for Cell Analyses and Modelling, University of Connecticut Health Centre,
Farmington, Connecticut 06032, USA. 3Experimental Medicine and Pharmacology, Centre for
Micro-vascular Research, William Harvey Research Institute, Barts and The London School of Medicine and
Dentistry, Queen Mary, University of London, Charterhouse Square, London, EC1M 6BQ, UK. 4
Molecu-lar Cytology, Swammerdam Institute for Life Sciences, University of Amsterdam, Amsterdam 1098XH,
The Netherlands. 5Department of Angiogenesis,Walter-Brendel-Center of Experimental Medicine
Nature Communications (2016); 7:10493
Niels Heemskerk
1, Lilian Schimmel
1,
Chantal Oort
1, Jos van Rijssel
1, Taofei
Yin
2, Bin Ma
3, Jakobus van Unen
4,
Bettina Pitter
5, Stephan Huveneers
1,
Joachim Goedhart
4, Yi Wu
2, Eloi
Montanez
5, Abigail Woodfin
3& Jaap D.
van Buul
1F-actin-rich contractile
endothelial pores prevent
vascular leakage during
leukocyte diapedesis
through local RhoA
signalling
Summary
During immune surveillance and inflammation, leukocytes exit the vascula-ture through transient openings in the endothelium without causing plasma leakage. However, the exact mechanisms behind this intriguing phenomen-on are still unknown. Here we report that maintenance of endothelial barrier integrity during leukocyte diapedesis requires local endothelial RhoA cycling. Endothelial RhoA depletion in vitro or Rho inhibition in vivo provokes neu-trophil-induced vascular leakage that manifests during the physical move-ment of neutrophils through the endothelial layer. Local RhoA activation ini-tiates the formation of contractile F-actin structures that surround emigrating neutrophils. These structures that surround neutrophil-induced endothelial pores prevent plasma leakage through actomyosin-based pore confinement. Mechanistically, we found that the initiation of RhoA activity involves ICAM-1 and the Rho GEFs Ect2 and LARG. In addition, regulation of actomyo-sin-based endothelial pore confinement involves ROCK2b, but not ROCK1. Thus, endothelial cells assemble RhoA-controlled contractile F-actin structu-res around endothelial postructu-res that prevent vascular leakage during leukocyte extravasation.
Introduction
The clinical signs of inflammation, redness, heat, swelling and pain are cau-sed by the acute inflammatory response including increacau-sed vasodilatation, enhanced microvascular permeability and leukocyte recruitment. During in-flammation the endothelial barrier becomes more permissive for large mo-lecules, leading to local plasma proteins leakage and oedema formation. Whether leukocyte transendothelial migration (TEM) directly causes incre-ased microvascular permeability has been controversial for decades. Certain studies suggested leukocyte adhesion and transmigration to be the critical events leading to tissue damage and organ failure during inflammation and ischemia reperfusion 1,2, since neutrophil depletion or CD11-/CD18-blocking
antibodies have been shown to attenuate vascular injury under these circum-stances 2–5. However, when microvascular permeability was measured
simul-taneously with leukocyte–endothelial interactions, local plasma leakage sites were often different from those of leukocyte adhesion or transmigration 6–11.
Recently, it has been shown that intravenous injection of tumour necrosis fac-tor (TNF)-α caused significant leukocyte adhesion and transmigration but did not affect basal microvessel permeability 12. Moreover, several studies have
shown that the timing of leukocyte adhesion and transmigration are not well linked with the evoked permeability change during acute inflammation 13–16.
Most of the abovementioned studies are descriptive, molecular evidence for the uncoupling between leukocyte TEM and vascular permeability has been recently shown by Wessel and colleagues. They mechanistically uncoupled leukocyte extravasation and vascular permeability by showing that opening of endothelial junctions in those distinct processes are controlled by different tyrosine residues of VE-cadherin in vivo 17. However, how the endothelium
maintains a tight barrier during leukocyte transendothelial migration is still unknown 18.
Here we investigate the mechanism by which endothelial cells (ECs) prevent vascular leakage during leukocyte TEM. We examine the correlation between neutrophil extravasation and the evoked permeability changes du-ring acute inflammation in vitro and in vivo. Spatiotemporal RhoA activation during leukocyte crossing is measured using a recently developed RhoA bi-osensor 19. In addition, we use fluorescently-tagged Lifeact and Lifeact-EG-FP transgenic knock-in mice to investigate endothelial filamentous (F)-actin dynamics in remodelling junctions during neutrophil diapedesis in vitro and
in vivo, respectively. We show that endothelial pore restriction limits vascular
leakage during leukocyte extravasation, which is driven by a basolateral ac-tomyosin-based structure that requires local endothelial RhoA activation.
Results
RhoA controls vascular leakage during leukocyte diapedesis
To investigate the molecular mechanism that controls endothelial barrier function during neutrophil TEM, we simultaneously measured neutrophil transmigration kinetics and fluorescein isothiocyanate (FITC)–dextran lea-kage across TNFα-stimulated human umbilical vein endothelial cells (ECs) towards a C5a gradient, for 60 min. Neutrophil transmigration across control ECs was associated with minimal FITC–dextran leakage (Figure 1A). Incre-asing neutrophil numbers in the upper compartment up to 10-fold did not induce FITC–dextran leakage, indicating that ECs maintained their barrier function, despite increased numbers of transmigrating neutrophils (Figure S1A). To investigate the functional role of RhoA in EC barrier maintenance during neutrophil TEM, we depleted RhoA using siRNA (Figure S1B). We found that endothelial RhoA depletion increased FITC–dextran leakage du-ring neutrophil extravasation, whereas minimal FITC–dextran leakage was measured during neutrophil crossing through control ECs (Figure 1A, B). Correlation analysis showed that the increase in FITC–dextran leakage was highly correlated to neutrophil transmigration (Figure 1C). Note that endothelial RhoA depletion did not alter FITC–dextran leakage under basal conditions, which was comparable to control EC (Figure 1A, B). Moreover, endothelial resistance measured under physiological flow conditions was significantly reduced during transmigration of neutrophils across Rho-inhi-bited endothelium (Figure S1C). We next investigated the role of RhoA in EC barrier maintenance during neutrophil TEM in vivo. Vessel permeability was monitored by Tetramethylrhodamine (TRITC)–dextran leakage into the cremaster of C57BL/6 wild type (WT) or LysM–GFP mice during interleukin (IL)-1β and TNF-α-stimulated neutrophil recruitment. Intrascrotal administra-tion of anti-PECAM-1 labelling antibody resulted in a strong labelling of EC junctions in cremasteric venules (Figure 1D). Administration of IL-1β and TNF-α enhanced leakage of intravenous TRITC–dextran into the interstiti-um and neutrophil recruitment into the cremaster (Figure 1D; Figure S1D). Rho inhibitor I (C3)-treated animals showed similar extravasated neutrophil levels, however, TRITC–dextran leakage in those animals was highly incre-ased compared with IL-1β and TNF-α administration alone (Figure 1D, E). Although no change in neutrophil extravasation was measured in the presen-ce or absenpresen-ce of C3, we cannot exclude that the inhibitor affects other presen-cells. Neutrophil extravasation and TRITC–dextran leakage in WT mice were not correlated in individual mice, although there was an overall association bet-ween extravasation and permeability, whereas the two processes in Rho-in-hibited animals showed a highly significant correlation (Figure 1F). Animals treated with C3 alone showed unaltered basal vascular permeability (Figure
3
S1E). Thus, neutrophil extravasation and evoked changes in vascular per-meability during inflammation are not correlated. However, when endothelial RhoA is inhibited, neutrophil diapedesis provokes vascular leakage, sugge-sting that endothelial RhoA is required to maintain a tight EC barrier during leukocyte diapedesis in vivo.
Spatiotemporal RhoA activation during leukocyte diapedesis
To examine spatiotemporal RhoA activation in ECs during EC barrier main-tenance associated with neutrophil TEM, we used a recently developed fluo-rescence resonance energy transfer (FRET)-based RhoA biosensors called the Dimerization Optimized Reporter for Activation (DORA) RhoA sensors (Figure 2A) 19. DORA RhoA biosensors design were based on the
publis-hed RhoA biosensor 20. The ON-state FRET efficiency of the GTPase was
improved through modelling of the fluorescent protein dimers and the GT-Pase-effector domain complexes. Stable α-helical repeats from ribosomal protein L9, rather than an unstructured linker, were inserted between the flu-orescent proteins to disrupt dimerization and diminish FRET efficiency in the inactive state (Figure 2A). As a control, DORA RhoA mutant Protein kinase N (PKN) was developed to report misalignment of Cerulean3 (Cer3) and Venus image before and after image registration, motion artefacts or pH changes affecting the sensors fluorescent proteins. Glutamine substitution for a leuci-ne at position 59 in the PKN domain prevents PKN binding to activated RhoA
21. The characterizations of both DORA RhoA biosensors are described in
online methods (Figure S2; Figure S3A). From these validation experiments, we conclude that the DORA RhoA biosensor accurately reports RhoA dyna-mics in ECs downstream from endogenous stimuli such as thrombin (Figure 2B; Figure S1F).
To study spatiotemporal RhoA activation in ECs during neutrophil TEM, we expressed the DORA RhoA biosensor in ECs and investigated RhoA activation following neutrophil extravasation under physiological flow conditions. Important to note, Venus and Cer3 emission were simultaneously recorded utilizing a double-camera system, since sequential image acquisi-tion resulted in moacquisi-tion artefacts induced by migrating leukocytes displacing fluorescent signals in ECs. We found unaltered RhoA biosensor activation during neutrophil rolling and crawling over the endothelium (Figure 2C, G). Also RhoA activation during the initial opening of EC junctions was found to be unaltered (Figure 2D). However, RhoA biosensor activity in the lium was locally increased at sites were neutrophils breeched the endothe-lium, between the first and second minute of neutrophil diapedesis (Figure 2D-F; Figure S2D; Movie S1). On the basis of the normalized ratiometric imaging and the relative displacement of the sensor, the data showed a 1.2-fold increase in FRET ratio on diapedesis, comparable to what has been ob-served during RhoA activation after thrombin stimulation (Figure 2H, I). The
Figure 1
a b c d e 0 20 40 0 100 200 300 400 500 Neutrophils / 5x104 m2 P=0.16 n.s. IL-1β/TNFα µ 0 20 40 0 200 400 600 + C3 P=0.006** IL-1β/TNFα Neutrophils / 5x104µm2 IL -1β/TNFα IL -1β/TNFα +C3PECAM-1 PMN TRITC-dextran Merge 0 2 4 6 8 10 12 14 16 18 20 0 2 4 6 8 10 12 Time (min) 0 2 4 6 8 10 12 14 16 18 20 0.8 0.9 1.0 1.1 1.2 1.3 1.4 1.5 1.6 Time (min) Transmigrated neutrophils (a.u)
Dextran
leakage (a.u) siCTRL
siCTRL + PMN siRhoA + PMN siRhoA siCTR L siRho A 1 2 3 4 5 6 7 8 9 10 11 12 13 T = 20min Tr an sm ig ra te d ne ut ro ph ils fo ld in cr ea se ns siCTR L siRho A siCTR L +PM N siRho A+PM N 1.0 1.1 1.2 1.3 1.4 1.5 1.6 T = 20min FI TC -d ex tr an 70 kD a le ak ag e fo ld in cr ea se *** ns ns 0 5 10 15 20 0.8 1.0 1.2 1.4 1.6 1.8 2.0 T = 5-20min Neutrophil diapedesis fold increase FI TC -d ex tr an 70 kD a le ak ag e fo ld in cr ea se 0 5 10 15 20 0.8 1.0 1.2 1.4 1.6 1.8 2.0 T = 5-20min Neutrophil diapedesis fold increase FI TC -d ex tr an 70 kD a le ak ag e fo ld in cr ea se f siCTRL siRhoA Extravascular TRITC-dextran fluorescence (a.u.) P=0.004** P=0.35 n.s. Extravascular TRITC-dextran fluorescence (a.u.) 0 10 20 30 40 50 Neutrophil extravasation C3 -- +- +- ++ IL-1β/TNFα Ex tr av as at ed ne ut ro ph ils (5 x1 0 4m 2) µ
Figure 1. Impaired endothelial RhoA function results in increased vascular lea-kage during leukocyte diapedesis in vivo
(A) Extravasation kinetics of calcein-red-labelled neutrophils and FITC–dextran through TNF-α treated ECs cultured on 3-μm pore permeable filtres. Neutrop-hils transmigrated towards a C5a chemotactic gradient in the lower compartment.
3
Four conditions were tested; RhoA depletion (EC)+neutrophils (Orange line), con-trol+neutrophils (purple line), RhoA depletion (EC) only (red line) and control only (green line). (B) Quantification of FITC–dextran and neutrophil extravasation after 20 min of neutrophil transmigration. Immunoblot of RhoA silencing can be found in (Figure S8). (C) Correlation analysis of dextran and neutrophil extravasation ki-netics through control and RhoA-depleted ECs. (D) Confocal intravital microscopy of 20–80 μm diameter cremasteric venulesin LysM–GFP mice (green neutrophils) immunostained in vivo for EC junctions by intrascrotal injections of fluorescent-la-belled PECAM-1 (blue) and stimulated for four hours with IL-1β and TNF-α only, or with Rho-inhibitor (C3). A second dose of Rho inhibitor was given intrascrotally and TRITC–dextran (40 kDa) was injected intravenously at T=2 h and allowed to circulate until T=4 h. Scale bar, 100 μm. (E) Neutrophil extravasation in animals left unstimulated (control), stimulated with C3 alone, IL-1β/TNFα treated, IL-1β/TNFα treated+C3 or IL-1β/TNFα treated+neutrophil depletion. (F) Correlation analysis of dextran and neutrophil extravasation kinetics in animals stimulated with IL-1β/TNFα alone or with IL-1β/TNFα treated+C3. ***P<0.001 control versus RhoA-depleted HUVEC (ANOVA) or P=0.3504 control versus RhoA-depleted HUVEC (Student’s t-test) (B). r=0.2547 P=0.359 (Pearson’s correlation) transmigrated neutrophils ver-sus FITC–dextran leakage in control HUVECs or r=0.6345 **P<0.01 (Pearson cor-relation) transmigrated neutrophils versus FITC–dextran leakage in RhoA-depleted HUVECs (C). P=0.4230 IL-1β/TNFα versus IL-1β/TNFα+C3 (Student’s t-test) (E). r=0.8258 **P<0.01 (Pearson’s correlation) transmigrated neutrophils versus FITC– dextran leakage in IL-1β/TNFα+Rho inhibitor treated mice (F). Data are from three experiments (A–C) or are representative of 5 to 13 (D–E) or 9 (F) independent expe-riments ((D–F) one mouse per experiment; error bars (A–C,E,F), s.e.m).
negative control DORA RhoA biosensor (mutant PKN) showed no change in FRET during leukocyte diapedesis (Figure S3B, C; Movie S2). Importantly, expressing the DORA RhoA biosensors in ECs did not interfere with neutrop-hil TEM. Thus, endothelial RhoA is transiently and locally activated during the final stage of neutrophil diapedesis, but not during crawling or opening of endothelial junctions, indicating a role for local RhoA activity in EC barrier maintenance during the final stage of neutrophil extravasation.
F-actin-rich endothelial pores during diapedesis
To investigate endothelial F-actin dynamics during neutrophil diapedesis, we transfected ECs with GFP- and/or mCherry-tagged Lifeact 22. It is
im-portant to note that phalloidin staining to visualize F-actin cannot be used to investigate endothelial actin structures that are in close proximity of transmi-grating leukocytes, since F-actin in both leukocytes and ECs are visualized by phalloidin staining, making it impossible to discriminate between the two (Figure S3D). Transmigrating neutrophils initiated small endothelial pores in
00:12 00:00 00:24 00:36 00:48 01:00 01:12 01:24 01:36 01:48 02:00 DIC Venus/Cer3 Merge - 00:12 d
DIC Venus/Cer3 Merge
Neutrophil diapedesis 00:12 -01:36 00:24 00:36 00:48 01:00 01:12 01:24 01:36 01:48 02:00 - 05:36
Figure 2.
f g -2 -1 0 1 2 3 4 5 6 0.9 1.0 1.1 1.2 1.3 1.4 1.5 1.6 Time (min) N or m al iz ed Ve nu s/ C er 3 em is si on ra tioDORA RhoA activation Neutrophil adhesion -2 -1 0 1 2 3 4 5 6 0.9 1.0 1.1 1.2 1.3 1.4 1.5 1.6 Time (min) N or m al iz ed Ve nu s/ C er 3 em is si on ra tio
DORA RhoA activation Neutrophil diapedesis DIC Venus/Cer3 Merge c
e DIC Venus/Cer3 Merge
Neutrophil adhesion
Neutrophil adhesion
Neutrophil diapedesis Neutrophil adhesion
DORA RhoA biosensor
DORA RhoA biosensor a
cpPKN RhoA
N dcpVen dCer3 C N cpPKN dcpVen dCer3 RhoA C
L59Q
DORA RhoA biosensor DORA RhoA mut PKN biosensor
L9 L9 L9 L9 L9 L9 b Flow Flow Venus/Cer3 Before + Thrombin
DORA RhoA biosensor
0 1 2 3 4 5 6 7 8 9 10 0.9 1.0 1.1 1.2 1.3 1.4
DORA RhoA activation Time (min) N or m al iz ed Venus/Cer3 em is si on r at io * i h
3
Figure 2. Spatiotemporal RhoA activation during neutrophil TEM
Endothelial RhoA is locally and transiently activated during neutrophil extravasa-tion (A) Schematic illustration of the DORA RhoA sensor design containing Rho effector PKN (red), circular permutated Venus (yellow), structured linker protein L9 (green), circular permutated Cer3 (blue) and RhoA GTPase (green),left panel. Right panel shows the DORA RhoA mutant PKN biosensor that was developed as a negative control biosensor, the glutamine was substituted for the leucine at position 59 in the PKN domain. This mutation prevents binding of PKN to activated RhoA. (B) Time-lapse Venus/Cer3 ratio images of DORA RhoA biosensor simultaneously recorded with an epi-fluorescent microscope showing spatiotemporal RhoA activa-tion upon thrombin treatment (1 U ml−1) in HUVECs. Filled arrows indicate RhoA activation. Scale bar, 10 μm. Calibration bar shows RhoA activation (Red) relative to basal RhoA activity (Blue). (C) Epi-fluorescent live-cell imaging of HUVEC ex-pressing the DORA RhoA biosensor during neutrophil adhesion under physiological flow conditions (0.8 dyne per cm2). Red open arrows indicate adherent neutrophils. Scale bar, 10 μm. Calibration bar shows RhoA activation (red) relative to basal RhoA activity (blue). (D) Epi-fluorescent live-cell imaging of HUVECs expressing the DORA RhoA biosensor during neutrophil TEM. Time-lapse images of DIC (upper) Venus/Cer3 ratio images of DORA RhoA biosensor (middle) and Merge (bottom) during leukocyte diapedesis. Open arrows indicates adherent neutrophil at the apical side of the endothelium. Filled arrows indicate local RhoA activation during neu-trophil diapedesis. Scale bar, 10 μm. (E) Detailed zoom of RhoA activation during neutrophil adhesion (open arrows) prior diapedesis at time point t=−00:12 min. (F) Detailed zoom of local RhoA activation during neutrophil transmigration at time point t=01:12 min. Filled arrows indicate local RhoA activation during neutrophil diapedesis. Scale bar, 10 μm. (G) Quantification of temporal RhoA activation during multiple neutrophil transmigration events starting at time zero (arrow). (H) Quan-tification of temporal RhoA activation during multiple neutrophil adhesion events starting at time zero (arrow). (I) Quantification of DORA RhoA biosensor activation after thrombin treatment (1 U ml−1) in HUVEC. Asterisk indicates thrombin additi-on. Data represent mean and s.e.m of 7 experiments (G) 5 experiments and (H) 10 experiments (I).
the endothelial lining. To study those endothelial pores at high resolution, transmigrating neutrophils were fixed with formaldehyde when partly bree-ched the endothelium. Confocal microscopy imaging and three-dimensio-nal (3D) reconstruction showed that ECs assembled F-actin-rich structures around endothelial pores through, which neutrophils transmigrated, both du-ring transcellular and paracellular migration (Figure 3A; Movie S3-5). Dudu-ring paracellular migration, the junctional protein VE-cadherin was distributed to the endothelial pore margins (Figure 3A). Interestingly, using ECs expressing either Lifeact-GFP or Lifeact-mCherry, we found that paracellular pores were
Lifeact
DIC VE-cad DAPI Merge
Paracellular
Transcellular
a
b Lifeact DAPI VE-cad Merge
Merge Lifeact-GFPmCherryLifeact Z-distance 4.8 µm 3.9 µm 3.0 µm 2.1 µm 1.2 µm 4.8 µm 1.2 µm Adhesion-stage Diapedesis-stage EC VE-cadherin complex F-actin X-Z (10 µm) Y-Z ( 10 µm) Y X e c d Basal Apical * * * * * * * * * * * * Flow Flow 0 20 40 60 80 100 F-ac tin ri ch en do th el ia lp or es % Paracellular Transcellular ns ** **
Neutro Mono CD3+ T-lympho
Figure 3. ECs assemble F-actin-rich ring-like structures around transmigrating leukocytes
(A) Confocal imaging of para- and transcellular migrating leukocytes through Life-act-GFP expressing HUVECs. Filled arrows indicate EC F-actin (Lifeact in green) assembly around extravasating leukocytes. Open arrows indicate VE-cadherin
(di-3
rectly labelled Alexa-647 antibody in red) distribution to the F-actin structure sites during paracellular diapedesis. Asterisks indicate extravasating leukocyte (DAPI in blue) in DIC. Flow speed 0.8 dyne per cm2. Scale bar, 5 μm. (B) Confocal imaging
showing a Z-stack of Lifeact-GFP and Lifeact-mCherry positive endothelial mem-brane structures from apical to basal plane. Open arrows indicate filopodia-like prot-rusions at the apical site of the structure. Filled arrows indicate the cortical actin ring at the basolateral site that appeared during leukocyte crossing. Asterisk indicates extravasating leukocyte (DAPI in blue). Scale bar, 5 μm. (C) Cartoon of endotheli-um during basal-stage and leukocytes diapedesis showing filopodia-like protrusions and the basolateral F-actin ring. (D) X–Z (10 μm) and Y–Z (10 μm) projections of confocal Z-stack shown in (Figure 1B). (E) Quantification of percentage F-actin-rich endothelial pores associated with neutrophil and monocyte extravasation during para- and transcellular migration. Statistical significance was tested with ANOVA. Data are representative for three independent experiments (A–E) with 37–65 trans-migration events per group (error bars (E) s.e.m).
formed by at least two ECs. At the structures apical site, filopodia-like prot-rusions were found, whereas at the basolateral site, a cortical actin ring ap-peared during leukocyte crossing (Figure 3B-D). In contrast to VE-cadherin distributed to the pores margins, the junctional protein PECAM-1 was locali-zed around the basolateral F-actin ring and distributed to apical protrusions surrounding migrating leukocytes during trans-and paracellular migration. (Figure S3E, F). Moreover, we found that the adhesion molecule ICAM-1 was localized in the apical protrusions at endothelial pores (Figure S4A). We found that ~90% of all neutrophils and monocytes used the paracellular route, whereas ~10% migrated transcellular, in line with the migratory prefe-rence for neutrophils and monocytes found in vivo 23 (Figure 3E). Note that all
these diapedesis events by either neutrophils or monocytes were associated with basolateral F-actin ring formation around endothelial pores (Figure 3E). Within the paracellular route of migration, leukocyte transmigration through a bi-cellular or a multicellular junction was ~50% (Figure S4B). In conclusi-on, ECs assemble F-actin-rich ring-like structures around endothelial pores through which neutrophils and monocytes transmigrate. This data indicate that maintenance of EC barrier function during leukocyte diapedesis invol-ves actin cytoskeleton strengthening around endothelial pores. Basolateral F-actin ring formation may tighten the endothelial barrier during neutrophil crossing, making the leukocyte-induced endothelial pore impermeable for macromolecules.
F-actin-rich endothelial pores are confined in size
Electron microscopy studies showed that ECs maintain intimate contact with transmigrating neutrophils during the entire transmigration process 15,24. To
c DIC Lifeact-GFP 00:20 00:00 00:40 01:00 01:20 01:40 02:00 02:20 - 00:20 02:40 b a -1 0 1 2 3 -5 0 5 10 15 20 Endothelial pore Neutrophil cell body (apical)
Time (min) -1 0 1 2 3 -100 -50 0 50 100 150 200 250 Endothelial pore Neutrophil cell body (apical) Neutrophil cell body (under EC)
Time (min) A re a µm 2 Neutrophil diapedesis Neutrophil diapedesis Flow 0 20 40 60 80 100 Endothelial pore size ns ns ns e Neutro
para Monopara transTcell
Area
µm
2
Endothelial pore size F-Actin rich pore Cell-Cell Junction d
Endothelial pore size
Adhe ring Early diape desis mid d iaped esis late d iaped esis 0 10 20 30 40 50 A re a µm 2 **** **** **** ns
adhesion early diapedesis mid diapedesis late diapedesis
Adhe sion Early diape desis mid d iaped esis late d iaped esis Adhe sion Early diape desis mid d iaped esis late d iaped esis 0 10 20 30 40 50 60 70 80 90 100 F-ac tin st ru ct ur es % F-actin ring Apical F-actin protrusions **** **** ns ns ns ns Adhe ring Early diape desis mid d iaped esis late d iaped esis Adhe ring Early diape desis mid d iaped esis late d iaped esis 0 25 50 75 100 125 A re a µm 2 Neutrophil cell body apical
Neutrophil cell body under EC ns **** **** **** **** **** A re a µm 2
Figure 4. Endothelial pores formed during para- and transcellular leukocyte transmigration are confined in size
(A) Epi-fluorescent live-cell imaging of ECs expressing Lifeact-GFP. Red open ar-rows indicate F-actin-rich endothelial pore formation during leukocyte diapedesis
3
under physiological flow conditions (0.8 dyne per cm2). Filled arrows indicate
extra-vasating leukocyte in DIC. Dashed lines indicates neutrophil localization under the endothelium. Scale bar, 10 μm. (B,C) Quantification of size changes occurring in the neutrophil cell body and endothelial pore during neutrophil diapedesis. Endothelial pore size (red), neutrophil cell body apical (blue) and neutrophil cell body under EC (yellow), diapedesis starts at time zero. (D) Quantification of Neutrophil size, F-actin-positive ring structures, F-actin positive apical protrusions and endothelial pore size. (E) Quantification of endothelial pore size for neutrophils and monocytes during paracellular migration. ****P<0.0001 (analysis of variance). Data are repre-sentative of four independent experiments (D,E) with 40 transmigration events per group. Data in B and C are representative of 10 transmigration events (error bars (B–E) s.e.m).
examine the dynamic contact between ECs and extravasating neutrophils, we examined F-actin-enriched endothelial pore shape and size in relation to neutrophil size. Real-time recordings of transmigrating neutrophils through ECs expressing GFP-tagged Lifeact showed increased F-actin assembly around endothelial pores (Movie S6).
The kinetics of neutrophil diapedesis is on an average 2 min and can be distinguished into early, mid and late diapedesis based on endothelial pore size and neutrophil morphology (Figure 4A-C). Endothelial pore for-mation started when neutrophils partly breeched the endothelium, defined as early diapedesis. Following neutrophil diapedesis, most endothelial po-res are maximal enlarged 1 min after transmigration was initiated, defined as mid diapedesis (Figure 4A-C). Subsequently, the endothelial pore is clo-sed in conjunction with transmigrating neutrophils until completely under the endothelium, a stage defined as late diapedesis (Figure 4A-C). Real-time imaging of neutrophil diapedesis under physiological flow conditions showed that neutrophil total surface area before TEM was roughly 100 μm2, which
was reduced to <20 μm2 to fit the confined gap in the endothelium that had
a maximal inner-surface area of 19 μm2 (Figure 4B,C). To investigate the
morphology of de novo formed F-actin-positive rings and F-actin-positive apical protrusions that surround endothelial pores during neutrophil TEM, we trapped neutrophils at different stages of diapedesis. Interestingly, de novo formed F-actin-positive rings surrounding endothelial pores were found throughout all diapedesis steps, but not during neutrophil adhesion or cra-wling steps (Figure 4D; Figure S4C). Quantification of endothelial pore size showed significant larger pores during mid diapedesis than during early and late diapedesis when pores open and close, respectively (Figure 4D). We next measured, the pore size width, length and height of F-actin-rich en-dothelial pores surrounding transmigrating neutrophils and monocytes. On average, endothelial pores are 4-μm wide, 6-μm in length and mostly oval shaped for all leukocytes migrating through the cell–cell junctions (Figure
a b
d
c
DIC Lifeact VEcadherin DAPI Merge
DIC Lifeact VEcadherin DAPI Merge
Basal
Apical
Basal
Apical
HUVEC treated with control siRNA
Endogenous RhoA depletion in HUVEC * * * * * * * Flow Flow e siCTR L siRho A siCTR L siRho A 0 10 20 30 40 50 60 70 80 90 100 F-ac tin st ru ct ur es % F-actin
ring protrusionsApical
ns **** early diape desis mid d iaped esis late d iaped esis early diape desis mid d iaped esis late d iaped esis 0 10 20 30 40 50 60 70 A re a µm 2 ****siRhoA siCTRL siCTR L siRho A 0 20 40 60 80 100 Tr an sm ig ra te d ce lls % ns siCTR L siRho A 0 20 40 60 80 100 A dh er en tc el ls # ns shCT RL shVE -cadh erin 0 10 20 30 40 50 60 Endothelial pore size A re a µm 2 ns
Figure 5. RhoA signalling is required for endothelial pore confinement
(A) Confocal imaging of paracellular migrating neutrophils through Lifeact-GFP ex-pressing HUVECs after 72-h transfection with control siRNA (upper panel) or RhoA siRNA (lower panel) under physiological flow conditions (0.8 dyne per cm2). Open
arrows and filled arrows indicate filopodia-like protrusions at the apical site and the cortical F-actin ring at the basolateral site of the endothelial pore, respectively. Asterisk indicates extravasating neutrophil (DAPI in blue). VE-cadherin (red). Scale bar, 5 μm. (B) Quantification of F-actin-positive ring structures and F-actin-positive apical protrusions in control versus RhoA-depleted ECs. (C) Quantification of endo-thelial pore size during early, mid and late diapedesis. (D) Quantification of
neutrop-3
hil adhesion and diapedesis through TNF-α treated ECs under physiological flow conditions after 72 h transfection with control siRNA (open bar) or RhoA siRNA (filled bar). (E) Quantification of endothelial pore size in control versus VE-cadherin depleted HUVECs. ****P<0.0001 (analysis of variance). Data are representative of four independent experiments (A–E) with >12 transmigration events per group (error bars (C, E) s.e.m).
S4D, F). In addition, we found that only during diapedesis ~40% of the endo-thelial pores contained F-actin-rich apical protrusions (Figure 4D). No such structures were detected during the crawling step. These structures reached a maximal height of 6–7 μm (Figure S4E). Transcellular pores were found to be more round or circular shaped and had an average circularity of about 1.3 according to the circularity index (Figure S4F). Endothelial pore sizes sho-wed remarkably little variation, despite leukocyte size, type or transmigration route (Figure 4E). Thus, endothelial pores induced by extravasating neutrop-hils and monocytes are confined in size and close directly behind transmigra-ted cells. Active endothelial pore confinement and pore closure corroboratransmigra-ted earlier findings that showed intimate contact between neutrophils and ECs during the entire TEM process and provides an explanation for limited tran-sendothelial escape of macromolecules during neutrophil crossing.
Pore confinement and pore closure requires endothelial RhoA
Our data showed that increased endothelial RhoA activity during neutrophil TEM corresponded to endothelial pore restriction and closure during mid and late diapedesis. To investigate whether RhoA regulates endothelial pore confinement, we silenced endothelial RhoA using siRNA. RhoA was succes-sfully depleted as shown by western blot analysis (Figure S5A). Confocal microscopy showed that RhoA depletion in ECs reduced Lifeact-GFP accu-mulation around endothelial pores, whereas Lifeact-GFP in the apical prot-rusions was still present (Figure 5A). Basal F-actin rings in RhoA-depleted ECs were significantly reduced compared to control conditions (Figure 5B). Endothelial RhoA depletion had no effect on the formation of F-actin-rich api-cal protrusions (Figure 5B). Quantification of endothelial pore size showed that in the absence of RhoA, endothelial pores were not only larger than en-dothelial pores formed in control ECs but also did not close properly (Figure 5C). Note that neutrophil adhesion and transmigration under physiological flow conditions were unaltered in RhoA-depleted ECs (Figure 5D). To study if VE-cadherin signalling regulates endothelial pore size, we depleted VE-ca-dherin and analysed endothelial pore size. However, VE-caVE-ca-dherin depletion had no effect on endothelial pore size (Figure 5E; Figure S5B-D). In conclu-sion, RhoA facilitates endothelial pore confinement and pore closure during leukocyte diapedesis.
a DIC Lifeact pMLC Merge
Distribution
Asymmetry=ROI IntensityROI Intensity - - BGBG F-actin in pore
Sum Intensity Z-slices
Cell-Cell Junction b * * Flow Neutro Mono 1.0 1.5 2.0 2.5 D is tr ib ut io n A sy m m et ry ns e PECAM-1 f pMLC PECAM-1 0 2 4 6 8 Cremasteric venules D is tr ib ut io n A sy m m et ry pMLC Merge Cremaster venules PECAM-1 Neutrophils Merge Isolectin B4 Lifeact Merge Isolectin B4 Lifeact Merge c ROI Retinal vasculature of lifeact-EGFP
mice Area 0 10 20 30 40 50 60 Endothelial pore size d g Basal Apical Basal ROI Area µm 2
Figure 6. Endothelial pore confinement is driven by actomyosin contractility (A) Immunofluorescence analyses of MLC phosphorylation during neutrophil trans-migration. Open and filled arrows indicate Lifeact-mCherry (red) and MLC phosp-horylation (green) localization, respectively, during neutrophil transmigration under
3
physiological flow conditions (0.8 dyne per cm2). Asterisk indicates extravasating
leukocyte in DIC. Scale bar, 10 μm. (B) Quantification of F-actin distribution in en-dothelial pores surrounding transmigrating neutrophils and monocytes. Maximum intensity projection of F-actin in the endothelial pore was used to quantify F-actin distribution surrounding transmigrating leukocytes. Distribution asymmetry is defin-ed by the ratio of region of interest ROI-1 and ROI-2 correctdefin-ed for background. Sca-le bar, 5 μm. (C) Confocal imaging of F-actin dynamics during leukocyte diapedesis in retina vasculature of Lifeact-EGFP C57BL6 mice. Filled arrows indicate the vas-culature of mice retina, highly expressing Lifeact-GFP. Zoom of ROI, open arrows indicate the Lifeact-EGFP (green)-positive endothelial pore, filled arrows indicate transmigrating neutrophil. Scale bar, 5 μm. (D) Quantification of endothelial pore size in retina vasculature. (E) Confocal imaging of PECAM-1 in cremasteric venules during TNF-α and IL-1β induced neutrophil recruitment. Open arrows indicate PE-CAM-1 positive endothelial pores that surround extravasating neutrophils (filled ar-rows). Scale bar, 20 μm. (F) IF analyses of MLC phosphorylation during neutrophil transmigration into the cremaster of C57BL6 mice. Filled and open arrows indicate phospo-MLC and PECAM-1 localization to endothelial pores, respectively. Scale bar, 5 μm. (G) Quantification of pMLC and PECAM-1 localization in endothelial pores. We quantified MLC phosphorylation defined as distribution asymmetry. The distribution asymmetry uses the intensity of one ROI versus another ROI as indica-ted in B. Because MLC may occur at different heights within the pore we used max projection for this analysis. Data are representative of three independent experiments (A–G) with >12 transmigration events per group (error bars (B,D,G) s.e.m).
Pore confinement is driven by actomyosin contractility
To investigate how RhoA regulates endothelial pore confinement during leukocyte diapedesis, we examined RhoA effector myosin II activation. To study myosin II activation we locally measured myosin light-chain (MLC) phosphorylation on position Thr18 and Ser19. Immunofluorescent staining of pMLC showed an asymmetric phosphorylation pattern in F-actin-rich en-dothelial pores surrounding transmigrating neutrophils (Figure 6A). MLC was particularly phosphorylated at cortical actin bundles as part of the F-actin ring (Figure 6A; Figure S5E). Note that the uropod of the neutrophil is positive for MLC phosphorylation, most likely to retract its tail during transmigration 25.
In contrast to local MLC phosphorylation in control ECs, endothelial pores in RhoA-deficient ECs were enlarged and negative for local phospho-MLC (Figure S5E). In addition, we quantified Lifeact-GFP distribution around en-dothelial pores and found asymmetric F-actin distribution around enen-dothelial pores, indicative of increased tension (Figure 6B). To corroborate our fin-dings in vivo, we studied F-actin localization during leukocyte diapedesis in retinal vasculature of Lifeact-EGFP-transgenic knock-in mice 26.
endot-Figure 7 CTRLLARG P11 5 ECT2 LARG + P11 5 LARG +ECT 2 0 10 20 30 40 50 60 E nd ot he lia lp or e si ze µm 2 **** ns contr ol LARG + ECT 2 contr ol LARG + ECT 2 0 10 20 30 40 50 60 70 80 90 100 F-ac tin st ru ct ur es % F-actin ring Apical protrusions ns ns
early mid late early mid late 0 10 20 30 40 50 60 E nd ot he lia lp or e si ze µm 2 ****** ns Diapedesis
control siLARG + ECT2
a b c
f Adhesion Early Diapedesis
EC
VE-cadherin complex F-actin
Mid Diapedesis Late Diapedesis
Destabilizing
cell-cell junctions Leukocyteprobing Endothelial poreconfinement Endothelial poreclosure
CTRL + PMN siLAR G/sh Ect2 + PMN 0 5 10 15 Tr an sm ig ra te d ne ut ro ph ils fo ld in cr ea se T= 20 min * CTRL siLAR G/sh Ect2 CTRL + PMN siLAR G/sh Ect2 + PMN 1.0 1.2 1.4 1.6 1.8 2.0 2.2 FI TC -D ex tr an 70 kD a Le ak ag e Fo ld In cr ea se T= 20 min * d siCTR L siICA M-1 siCTR L +PM N siICA M-1 + PMN 1.0 1.1 1.2 1.3 1.4 1.5 1.6 FI TC -D ex tr an 70 kD a Le ak ag e Fo ld In cr ea se T= 20 min * siCTR L +PM N siICA M-1 + PMN 0 5 10 15 20 Tr an sm ig ra te d ne ut ro ph ils fo ld in cr ea se T= 20 min ** e
Figure 7. ICAM-1 regulates endothelial pore confinement through recruitment of the Rho GEFs LARG and Ect2
(A) Quantification of endothelial pore size in LARG, p115RhoGEF or Ect2 depleted ECs. (B) Quantification of endothelial pore size during early, mid and late diapedesis in control versus LARG+Ect2 depleted ECs. (C) Quantification of F-actin-positive ring structures and F-actin-positive apical protrusions in control versus LARG+Ect2 depleted ECs. (D) Quantification of FITC–dextran and neutrophil extravasati-on after 20 min of neutrophil transmigratiextravasati-on through cextravasati-ontrol and Ect2/LARG (D)
3
or ICAM-1-deficient ECs (E). (F) Model of endothelial pore formation. Adherent leukocytes destabilizing cell–cell junctions subsequently insert small pseudopodia between transient openings in the endothelium (leukocyte probing) during early diapedesis. The next step (mid diapedesis) involves local and transient RhoA ac-tivation that mediates endothelial pore confinement through the formation of a de novo basolateral F-actin ring and actomyosin contractility. Finally, persistent ac-tomyosin contractility closes the endothelial pore behind transmigrating leukocytes. ICAM-1 is involved in the regulation of endothelial pore confinement through re-cruitment of the Rho GEFs LARG and Ect2. Basolateral F-actin ring formation and actomyosin contractility tightens the endothelial barrier during leukocyte diapedesis, making the leukocyte-induced endothelial pore impermeable for macromolecules. ****P<0.0001 (ANOVA) (A–C).**P<0.01 (ANOVA) (B) *P<0.05 (ANOVA) (E, D).*P<0.05 control versus siLARG/shEct2 (D) **P<0.01 control versus siICAM-1 (E) (Student’s t-test). Data are representative of three independent experiments (A– E) with >12 transmigration events per group (A–C) (error bars (A–E) s.e.m). ANO-VA, analysis of variance.
helium and this allowed us to properly visualize F-actin in ECs in situ 26,27. We
found that endothelial pores induced by transmigrating neutrophils (isolectin B4-positive 28) were surrounded by Lifeact-EGFP-positive rings in retinal ECs
(Figure 6C). Quantification of these rings showed that endothelial pore size
in vivo was comparable to endothelial pores measured in the in vitro set-up
(compare Figures 6D and 4E). Lifeact was present in the basolateral ring and in apical protrusions that surrounded transmigrating neutrophils (Figure S5F). These data showed that apical membrane protrusions in vivo are rich for F-actin and surround adherent leukocytes. Next, we examined local MLC phosphorylation in WT mice during IL-1β and TNF-α-induced neutrophil re-cruitment in cremasteric venules. PECAM-1 was used as a marker to visua-lize endothelial pores in vivo 23 (Figure 6E). In line with our in vitro findings,
endothelial pores in mouse cremaster venules showed asymmetric MLC phosphorylation (Figure 6F, G). To address the role of ROCK in endothelial pore confinement we depleted the ROCK isoforms ROCK1 and ROCK2b in endothelial cells and examined vascular permeability during neutrophil dia-pedesis. In line with RhoA inhibition, silencing ROCK 1 and ROCK2b did not prevent the adhesion or transmigration of neutrophils through the endothe-lial monolayer (Figure 6A, B). Basal endotheendothe-lial barrier function in ROCK1 or ROCK2b deficient ECs was not affected. However, neutrophil diapedesis through ROCK2b, but not ROCK1 deficient ECs elicited increased endothe-lial permeability up to a twofold (Figure S6A, B). These findings may indica-te that endothelial pore confinement is mediaindica-ted through ROCK2b but not ROCK1. Altogether, local accumulation of F-actin and MLC phosphorylation is associated with neutrophil diapedesis in vitro and in vivo, suggesting that endothelial pore confinement is driven by local actomyosin contractility.
Pore confinement requires ICAM-1, LARG and Ect2
To investigate the signalling events upstream from RhoA, we focused on the involvement of guanine-nucleotide exchange factors (GEF) and performed a GEF screen that included: p115RhoGEF, Ect2 and LARG. Depletion of endothelial LARG together with Ect2 increased endothelial pore size, whe-reas depletion of LARG, p115 and Ect2 alone was not sufficient (Figure 7A; Figure S6C). Quantification of endothelial pore size at different stages of diapedesis showed that endothelial pores in LARG- and Ect2-depleted en-dothelium were enlarged during early and mid diapedesis, but had no effect on endothelial pore closure (Figure 7B). Under these conditions the number of F-actin-positive rings and F-actin-positive apical protrusions was unal-tered (Figure 7C). Enlarged endothelial pores in Ect2- and LARG-deficient ECs showed increased endothelial permeability during neutrophil diapede-sis, whereas basal EC barrier function was not affected (Figure 7D; Figure S6D-F). Neutrophil diapedesis through Ect2- and LARG-deficient ECs was slightly reduced (Figure 7D; Figure S6D). To study LARG and Ect2 recruit-ment to the intracellular tail of PECAM-1 or ICAM-1 we performed clustering experiments induced by anti-ICAM-1- or anti-PECAM-1-coated beads. We found that LARG and Ect2 are recruited to the intracellular tail of ICAM-1 (Figure S6G). Whereas PECAM-1 recruited only LARG, but not Ect2 to its intracellular tail on clustering (Figure S6H). To investigate whether ICAM-1 or PECAM-1 initiate and coordinate local RhoA activation and endothelial pore confinement during neutrophil diapedesis, we depleted ICAM-1 and PECAM-1 in ECs and examined the extravasation of calcein-red-labelled neutrophils and FITC–dextran across EC and measured endothelial pore size. We found that neutrophil transmigration through ICAM-1-deficient ECs compromised the endothelial barrier (Figure 7E; Figure S7A-C), whereas PECAM-1 depletion did not alter endothelial pore size or vascular leakage (Figure S7D, E). ICAM-1 and PECAM-1 depletion alone had no effect on endothelial permeability (Figure 7E; Figure S7D). In agreement with the exis-ting literature, endothelial ICAM-1 depletion significantly reduced the number of transmigrated neutrophils (Figure 7E). Neutrophil diapedesis through PE-CAM-1-deficient ECs showed no reduction in transmigration numbers (Figu-re S7D). These data point out an important role for ICAM-1, Ect2 and LARG signalling in controlling RhoA-mediated endothelial pore confinement and EC barrier protection during neutrophil diapedesis.
3
Discussion
Leukocytes that cross the endothelium induce large endothelial gaps without provoking leakage of plasma into the underlying tissue. However, the me-chanisms behind this intriguing phenomenon are unclear. The present stu-dy shows how ECs limit vascular leakage during leukocyte TEM. We found that RhoA-mediated F-actin rings contribute to endothelial pore confinement that maintains endothelial barrier integrity during leukocyte diapedesis. Neu-trophil diapedesis through ICAM-1-, Ect2/LARG- and RhoA-deficient ECs provokes vascular leakage that was highly correlated with neutrophil bree-ching events. Mechanistically, we found that the initiation of RhoA activity involves ICAM-1 and the Rho GEFs Ect2 and LARG. In addition, we found that the regulation of actomyosin-based endothelial pore confinement invol-ves ROCK2b, but not ROCK1. Our work identifies a novel mechanism that maintains endothelial barrier integrity during leukocyte extravasation, which is driven by a basolateral actomyosin-based structure that requires spatio-temporal RhoA cycling (Figure 7F).
Inflammation-driven leukocyte recruitment and vascular permeability are separable events 7,8,17. In line with these observations, we discovered that
during the TEM process endothelial RhoA plays a central role in EC barrier maintenance, but is redundant for leukocyte transmigration. In agreement with previous reports, blocking RhoA activity or depleting RhoA in ECs did not perturb adhesion 29 or transmigration 30. In contrast to the general
con-cept that RhoA activation is required for leukocyte adhesion and opening of endothelial junctions 31–35, we found that endothelial RhoA was locally and
transiently activated during the diapedesis step and not during neutrophil crawling, firm adhesion or opening of endothelial junctions prior to leukocy-te extravasation. These processes require a separaleukocy-te, RhoA-independent mechanism that allows leukocyte–EC adhesion or opening of endothelial junctions. In agreement with our findings, for both transmigration routes, en-dothelial pore opening is in part mediated by mechanical forces that are ge-nerated by migrating leukocytes. Polarized actin polymerization in the leuko-cyte elicits pulling and pushing forces that support the movement of immune cells through the confined endothelial pore 36,37. ICAM-1 is known to mediate
leukocyte–EC interactions, and crosslinking ICAM-1 using ICAM-1-coated beads or ICAM-1 antibodies results in increased RhoA activation suggesting a role for ICAM-1-mediated RhoA activation in leukocyte adhesion 32,38–41.
However, based on the spatiotemporal activation of RhoA, we suggest that ICAM-1-mediated RhoA signalling specifically occurs during the diapedesis step, in agreement with our data that shows ICAM-1 enrichment only at dia-pedesis sites. PECAM-1 was also detected at sites of diadia-pedesis, for either paracellular or transcellular migration. Recently, mechanical tension exerted
on ICAM-1 and also PECAM-1 enhanced RhoA activation and MLC phosp-horylation in ECs that was dependent on the recruitment of the RhoGEF LARG and ICAM-1 clustering 41,42. Our work shows that ICAM-1 clustering
in-deed promotes the recruitment of LARG and we additionally found Ect2 to be recruited on ICAM-1 clustering. Depletion of ICAM-1- and Ect2/LARG in ECs compromised the endothelial barrier during neutrophil diapedesis. Indicating that the ICAM-1-LARG/Ect2 signalling axis is likely to be activated upstream from RhoA activation, to regulate de novo F-actin rearrangements, endothe-lial confinement and barrier protection during neutrophil crossing. Altogether these data suggest that leukocytes exert mechanical forces on endothelial adhesion molecules that modulate endothelial F-actin cytoskeleton through mechanotransduction that may cause endothelial confinement.
The relationship between Ect2 and actomyosin contractility has been clearly established by several studies. For instance, Ect2 has been descri-bed to be involved in RhoA activation and contractile ring formation during cytokinesis 43. In addition, it has been shown that the molecular pathways
that regulates local RhoA activation during cytokinesis are also used to con-trol RhoA dynamics at the zonula adherens in interphase cells 44. Although no
proof for a role of Ect2 in endothelial junction regulation has been described we can speculate that Ect2 mediates similar functions in ECs, regulating ac-tomyosin contractility around the pore. The latter is probable, since depletion of LARG and Ect2 simultaneously results in larger pores without affecting the number of F-actin rings. In agreement with this hypothesis overall endothelial pore size in RhoA-deficient endothelial cells was increased due to the lack of basal F-actin ring formation. In addition, we observed that RhoA-deficient endothelial cells were unable to phosphorylate MLC near endothelial po-res. Surprisingly, the length-to-width aspect ratio between paracellular and transcellular pores was found to be different; however, this was not due a difference in nuclear size, shape or composition between neutrophils and monocytes. We speculate that in case of paracellular migration the amount of VE-cadherin disassembly at the pores margins may regulate pore size, which may affect the length/width ratio or circularity. In case of transcellular migration mechanical forces from the endothelium might counteract leukocy-te-induced forces from all directions forcing a circular passageway. A physical explanation for circular transcellular passages may also explained by cellular dewetting 45. Despite different length-to-width ratio of the pores, the overall
pore size is constant independent of leukocyte type, or transmigration route. This may indicate that the contractile forces generated during endothelial pore formation are high enough to counteract the mechanical forces gene-rated by transmigrating leukocytes. Alternatively, the RhoA-induced basola-teral F-actin ring itself may also add as a limiting factor for pore confinement on top of the actomyosin-based contractility. Endothelial pore confinement is probably not restricted to neutrophil and monocyte diapedesis but may also
3
occur during the diapedesis of other immune cells such as T-lymphocytes. Additional research is required to investigate this hypothesis. Surprisingly, we found that VE-cadherin depletion did not affect endothelial pore integrity, despite the prominent VE-cadherin localization at the pores margins. It is known that other junctional molecules such as N-cadherin may take over the function of VE-cadherin when VE-cadherin is depleted 46. The fact that
we see unaltered pore morphology in VE-cadherin deficient cells makes it conceivable that other molecules like N-cadherin take over the function of VE-cadherin controlling the integrity of the endothelial pore at its margins. It is evident that VE-cadherin plays a dominant role in endothelial barrier formation and regulation of leukocyte traffic through the endothelial barrier. For instance, locking VE-cadherin junctions reduces the number emigrating leukocytes 47 and the phosphorylation of VE-cadherin at Y731 induced by
adherent leukocytes prior diapedesis is a necessity for junctional destabili-zation and paracellular diapedesis 17. We cannot exclude a supportive role
for VE-cadherin in endothelial pore integrity, but we can exclude a direct role for VE-cadherin as a signalling molecule being involved in controlling and coordinating of endothelial pore confinement. Whether other junctional proteins such as JAM-A or CD99, that act distally from ICAM-1, and signal to RhoA to prevent leakage is currently unknown 23,48. We found that many
F-actin rings comprise apical membrane protrusions. These projections, also known as ‘docking structures’ or ‘transmigratory cups’, have been suggested to anchor endothelial adhesion receptors and therefore control leukocyte ad-hesion 30,39,49–52. However, the biological function of these structures is still
under debate. Interestingly, we found F-actin rings associated with leukocyte diapedesis that contained no apical protrusions suggesting that directional neutrophil diapedesis can occur through other mechanisms than ‘apical pro-jection’-guidance for instance transendothelial migration-promoting endot-helial chemokines that are locally released within the endotendot-helial pore 53.
In agreement with studies showing that apical projection assembly requires RhoG and Rac1 but not RhoA activity 39,54, we still observed apical
membra-ne protrusions around migrating leukocytes upon RhoA depletion, whereas the F-actin rings were significantly decreased. Suggesting that the basolate-ral F-actin ring and not the apical protrusions in the endothelial pore contri-bute to vascular leakage prevention during TEM. Interestingly, in drosophila, similar actomyosin networks have been found to rapidly close multicellular wounds by actomyosin contraction 55. Studies that investigated the
mecha-nisms by which ECs repair gaps in the endothelial monolayer, show that mechanical induced microwounds in the endothelium are healed by ventral lamellipodia, a mechanism that may also be involved in the closure of leuko-cyte-induced endothelial pores 56. Our data show that RhoA-mediated
con-tractile force generation responsible for endothelial pore restriction precedes ventral lamellipodia formation. Moreover, RhoA-mediated pore constriction
in ECs seems to be specific for the closure of leukocyte-induced endothelial pores, whereas ventral lamellipodia are also observed in maintenance of basal junctional integrity 57. On the basis of electron microscopy studies, it
has been suggested that the intimate contact between neutrophils and ECs during the entire transendothelial migration process limits leakage of plasma proteins. Moreover, several studies showed that ECs reseal the endothelial barrier before or in conjunction with neutrophils penetrating the basal lamina
15,58. In agreement with these ultrastructural studies we found that endothelial
pores closed before or in conjunction with neutrophils that fully breeched the endothelial lining. In the context of EC barrier maintenance it is well concei-vable that endothelial pore confinement and closure directly prevents vascu-lar leakage during leukocyte diapedesis whereas ventral lamellipodia restore junctional homeostasis after leukocyte crossing. Endothelial LSP1 has been implicated in a role for ‘dome’ formation and controlling permeability during TEM 58 and has been found to be activated downstream from ICAM-1
clus-tering 59. Altogether, this may open up the possibility that ICAM-1 clustering
activates RhoA through LSP1. However, future experiments should show if this signalling axis indeed is operational during TEM.
In conclusion, we have discovered that local RhoA-mediated F-actin rings in the endothelial lining prevent vascular leakage during leukocyte dia-pedesis. Elucidating the molecular and cellular mechanisms of barrier main-tenance during leukocyte diapedesis may have implications for the develop-ment of new therapies to restore normal homeostatic junctional remodelling to counteract vascular dysfunction during chronic inflammation.
3
Experimental procedures
DNA and RNA constructs
The DORA RhoA and DORA RhoA mutant PKN biosensors were a very kind gift of Yi Wu (University of Connecticut Health centre, Farmington, USA). Briefly, circular permutated PKN effector of RhoA coupled to dimeric circu-lar permutated Venus is linked via a ribosomal protein-based linker (L9H) with dimeric Cerulean3 (Cer3) coupled to RhoA. The DORA RhoA sequen-ce within a pTriEx-HisMyc backbone is cpPKN(S69-H97-GSG-S14-R68)-KpnI-GS-dcpVen-L9Hx3-BamHI-dCer3(G229)-NheI-RhoA-WT-HindIII. The DORA RhoA mutant PKN sequence within a pTriEx-HisMyc backbone is cpPKN (S69-H97-GSG-S14-R68, L59Q)-KpnI-GS-dcpVen-L9Hx3-BamHI-dCer3(G229)-NheI-RhoA-WT-HindIII. The Leucine (L) on position 59 in the PKN domain of the RhoA control biosensor is substituted for a glutamine (Q). The H1R, p63-RFP and mRFP-RhoGDI-α (pcDNA 3.1) were a kind gift of Joachim Goedhart (Swammerdam Institute for Life Sciences, University of Amsterdam, Amsterdam, the Netherlands). pLenti-Lifeact-mCherry, pLen-ti-Lifeact-GFP, were a kind gift of Stephan Huveneers (Sanquin, Amsterdam, the Netherlands). shRNA in pLKO.1 targeting VE-cadherin (12) B6 (TRCN 54090), GEF-H1 (TRCN 3174), GEF-H1 (TRCN 3175), p115RhoGEF (TRCN 33567), and Ect2 (TRCN 47686) were purchased from Sigma Aldrich missi-on library. siRNA targeting RhoA (sc-29471) (working cmissi-oncentratimissi-on 50 nM), ICAM-1 sc-29354 (50 nM), PECAM-1 sc-29445 (50 nM), LARG sc-41800 (50 nM), Rock-1 (sc-29473) (50 nM), Rock-2b (sc-29474) (50 nM),and scram-bled non-silencing siRNA were purchased from (Santa Cruz Biotechnology, Santa Cruz, CA).
Antibodies
Rabbit antibody against GEF-H1 (55B6) (Cat #4076) (1:1000 for WB), phosp-hor-Myosin light-chain Thr18/Ser19 (Cat #3674) (1:100 for IF), p115RhoGEF (D25D2) (Cat #3669) (1:1,000 for WB), RhoA (67B9) (Cat #2117X) (1:1,000 for WB) and CD31 (PECAM-1) (89C2) (Cat #3528) (1:1000 for WB) were purchased from Cell Signaling (BIOKE). Polyclonal rabbit antibody against Ect2 (Cat# 07-1364) (1:1,000 for WB) was purchased from Millipore. Po-lyclonal goat antibody against LARG (Cat#AF4737) (1:1,000 for WB) was purchased from R&D systems. Alexa Fluor 405 Phalloidin (1:100 for IF) was purchased from Promokine (Cat# PK-PF405-7-01). Polyclonal goat antibo-dy against VE-cadherin (C-19) (Cat# SC-6458) (1:1,000 for WB), Rock-2 (C-20) sc-1851 (1:1,000 for WB), Rock-1 (H-85) sc-5560 (1:1,000 for WB) were purchased from Santa Cruz (Bio-Connect). Polyclonal rabbit antibo-dy against ICAM-1 (Cat #SC-7891) (1:1,000 for WB) was purchased from Santa Cruz Biotechnology. Monoclonal mouse antibody against Filamin A
(Cat #MCA464S) (1:1,000 for WB) was purchased from Serotec. Monoclonal mouse Alexa Fluor 647 VE-cadherin (55-7H1) ( Cat# 560411) (1:100 for IF) and Alexa Fluor 488 PECAM-1 (Cat# 555445) (1:100 for IF) were purchased from Becton Dickinson. Monoclonal mouse antibody against Actin (AC-40) (Cat# A3853) (1:1,000 for WB) was purchased from Sigma. The Alexa Fluor 405 goat anti-rabbit IgG (Cat# A31556) (1:100 for IF), Alexa Fluor 647 chic-ken anti-goat IgG (Cat# A21469) (1:100 for IF), Alexa Fluor 488 chicchic-ken an-ti-rabbit IgG (Cat# A21441) (1:100 for IF) and Texas red 568 Phalloidin (Cat #T7471) (1:100 for IF) were purchased from Invitrogen. Secondary HRP-con-jugated goat anti-mouse, swine anti-rabbit antibodies (1:3,000 for WB) were purchased from Dako (Heverlee, Belgium). Hoechst 33342 (H-1399) (1:50 for IF) was purchased from Molecular probes (Life Technologies). All antibo-dies were used according to manufacturer’s protocol.
Cell cultures and treatments
Pooled human umbilical vein ECs (HUVECs) purchased from Lonza (P938, Cat # C2519A), were cultured on fibronectin (FN)-coated dishes in EGM-2 medium, supplemented with singlequots (Lonza, Verviers, Belgium) HU-VECs were cultured until passage 9. HEK-293T were maintained in DMEM (Invitrogen, Breda, The Netherlands) containing 10% (v/v) heat-inactivated fetal calf serum (Invitrogen, Breda, The Netherlands), 300 mg ml−1 L-gluta-mine, 100 U ml−1 penicillin and streptomycin and 1 × sodium pyruvate (In-vitrogen, Breda, The Netherlands). HeLa cells (American Tissue Culture Collection: Manassas, VA, USA) were cultured using DMEM supplied with Glutamax, 10% fetal bovine serum, Penicillin (100 U ml−1) and Streptomy-cin (100 μg ml−1). Cells were cultured at 37 °C and 5% CO2. HUVECs were treated with 1 U ml−1 thrombin (Sigma-Aldrich, St Louis, USA) for periods as indicated, pretreated with 10 ng/ml recombinant TNF-α (PeproTech, Rocky Hill, NJ) 24 h before each leukocyte TEM experiment, For Rho inhibition cells were preincubated with cell-permeable Rho inhibitor I (C3) (Cytoskele-ton, Cat# CT04) for 3 h. Cells were transfected with the expression vectors according to the manufacturer’s protocol with Trans IT-LT1 (Myrus, Madis-on, WI, USA). Lentiviral constructs were packaged into lentivirus in Human embryonic kidney (HEK)-293T cells by means of third generation lentiviral packaging plasmids (Dull et al., 1998; Hope et al. 1990). Lentivirus contai-ning supernatant was collected on day 2 and 3 after transfection. Lentivirus was concentrated by Lenti-X concentrator (Clontech, Cat# 631232). Trans-duced target cells were used for assays after 72 h. Cells were transfected with siRNA according to manufacturer’s protocol using INTERFERin (Po-lyplus). HeLa cell were transfected with Lipofectamine and imaged the next day. HeLa cells were treated with 100 μM histamine (Sigma-Aldrich, St Louis, USA) and 10 μM Pyrilamine (mepyramine; Sigma-Aldrich, St Louis, USA) for periods as indicated.
